When Does a Reorder Need Fresh Sampling Instead?

The most common question in a repeat programme is whether the new run needs a sample. The honest answer is that most do not, and that the ones which do are obvious once the five triggers are written down. Guessing costs either a sample nobody needed or a run nobody approved.

The distinction that matters is between a re-sample, which creates a new approval and a new reference, and a confirmation sample, which proves that an unchanged style can still be made. They cost roughly the same in calendar time and they do very different jobs.

Deciding which one is needed is mostly a matter of asking what would change if the sample failed. If the answer is a new approval, a new reference and a new revision, it is a re-sample. If the answer is that the run would be corrected and made again to the same revision, it is a confirmation sample.

What Does Sampling Actually Establish?

Sampling establishes that a specified product can be made, to a recorded measurement, from the materials named, with the construction described. It produces the approved sample, the reference measurements and, where relevant, the first test reports.

That is also why a first order and a reorder are different commercial events. The first buys the measurements, the reference and the reports; the repeat buys the same bag again, which is a much shorter document and a much shorter conversation when the file has been kept.

Everything a reorder needs flows from that. A repeat that changes none of those four inherits the approval and needs nothing new. A repeat that changes one of them needs the changed part re-established, which is what a re-sample does.

Which Five Triggers Force a Fresh Sample?

A component is discontinued and no equivalent exists. A tolerance is widened or a measured property changes. The construction, assembly or pattern changes. More than the agreed period has passed since the last run. Or a tested or regulated claim has to be re-established.

What Happens When a Component Is Discontinued?

First establish whether an equivalent exists. Equivalence means matching on every property that has a band, not on the nominal size. Where an equivalent exists, the swap goes through the change log with a comparison, and a re-sample is needed only if fit, assembly or a tested property is affected.

Where no equivalent exists, a re-sample is unavoidable, and it should be booked as such rather than absorbed into the run. A substitute discovered during production is the most expensive version of this problem, because by then the choice is between stopping the line and shipping something nobody approved.

There is a middle case worth naming: the component is discontinued but a successor code exists from the same maker. Where the maker states that the successor replaces the registered code and the properties match, most programmes accept it with a comparison and a register update rather than a full re-sample.

Where the successor changes anything that has a band, treat it as a substitution rather than a continuation. The word successor describes a catalogue relationship, not a technical one, and a buyer should not inherit a change in plating tone or pull weight because a maker reorganised its range.

When Does a Tolerance Change Force Re-Sampling?

When the band moves outward. Widening a band is an admission that the process cannot hold the original figure, and it changes what the product is. That deserves a sample made to the new band and a new reference, rather than a note on a file.

Tightening a band does not strictly require a sample, because the product is unchanged and only the control is stricter. It does require evidence that the process can hold the tighter figure, which usually means measuring ten units from the next run rather than making anything new.

There is a third case: a band that has never actually been measured. Where a programme discovers a property with a band nobody has tested, the honest step is to measure it on the next run rather than to re-sample, and then to tighten or widen the band on evidence rather than on assumption.

How Does Time Force a Fresh Sample?

Two clocks run, and only one of them is measured by the calendar. The retained reference ages steadily whether or not the style is made; the line and the shift pattern change whether or not anyone notices. Of the two, the second is what a confirmation sample actually tests, and it is the one most often overlooked when a gap is judged by date alone.

Two clocks run. The retained reference ages, and the people and machines change. A programme that has not run for a year has both problems at once: the gold sample is a year older and the line has made a hundred other styles since.

A confirmation sample resolves both in one step. It is made to the existing revision, from the existing register, and compared against the existing reference. Where it matches, the programme continues; where it does not, the question becomes which of the two clocks moved.

The two clocks also argue for a shorter period on a complex style. A bag with a welded seam, a moulded panel or a printed panel has more that can move in a year than a simple sewn style, and eight or nine months is a more honest interval for those than twelve.

What Is a Confirmation Sample?

A sample made to an unchanged specification to prove that the style can still be made. It creates no new approval, changes no reference and produces no new tolerance - it produces a comparison sheet and, usually, relief.

Purpose

Re-sample: new approval. Confirmation: proof the style is still makeable

Specification

Re-sample: changed. Confirmation: unchanged, same revision

Outcome

Re-sample: new reference and bands. Confirmation: a comparison sheet

Reference

Re-sample: a new gold sample. Confirmation: the existing one

Calendar

Both 6-10 working days, 12-15 for a complex style

When

Re-sample: a trigger fired. Confirmation: a long gap has passed

How Long Does Re-Sampling Take?

Sampling is 6-10 working days from a signed-off tech pack, and 12-15 working days for a complex style. Bulk production is 35-50 days from written sample approval. Those are the two windows, and a re-sample adds the first to the front of the second.

Where a laboratory is involved, its time is additional and outside both windows. That is worth stating plainly at the booking stage, because a laboratory queue is the one part of a re-sample schedule that neither side controls.

What Is Compared in a Re-Sample?

Everything with a band, plus whatever triggered the sample. A discontinued component means the substitute is compared against the registered item and the assembled bag is compared against the gold sample. A construction change means the affected seam or panel is measured and, where relevant, pull-tested.

The result is a new comparison sheet and, where the sample is approved, a new gold sample and a new revision of the spec file. Without those three, the re-sample has not actually changed anything and the next run will be judged against the old approval.

How Does a Re-Sample Change the Spec File?

The revision number moves, the changed field is edited, the revision log records the trigger, and the new gold sample is recorded with the run it came from. Four steps, the same four as any other change, because a re-sample is a change that happened to require a physical unit.

One detail is worth protecting: make the re-sample on the production line rather than in a sample room wherever the style is complex. A sample-room unit is made by the people who know the style, and it will pass a comparison that a production unit from the same setting might fail.

Finally, decide who signs the re-sample approval and put the name on the booking document. Where the sample is approved by someone without the authority to change the file, the revision never moves, and the next run is still booked against the old approval - which is the failure mode this whole exercise exists to prevent.

Where the re-sample was triggered by a discontinued component, the register line changes too, and the retained sample attached to it is replaced. Keeping the old retained piece, marked as superseded, is what lets a later run be traced back through the substitution.

When Is a Reorder Actually a New Product?

When the question is no longer whether it matches but what it is. A different construction, a different material family, a different capacity, or a different performance claim are all new-product decisions wearing a reorder's clothing.

The test is simple: if the approved comparison sheet can still be used to judge the result, it is a reorder. If it cannot, because the thing being measured no longer exists, it is a new product.

Where the answer is genuinely unclear, the economical way to settle it is to make the confirmation sample and read the comparison. A sample made to the existing revision answers the question with evidence, while a discussion about whether one is needed answers it with opinions and usually takes longer.

A different material family, a different construction, a different capacity or a new performance claim are all new-product decisions, and each one deserves the full development route.

One more consideration belongs here: a new product also resets the evidence. Test reports, certificates and any claim built on them were established for the old product, and a change that makes the comparison sheet unusable usually makes those reports unusable with it.

A useful way to set the interval is to tie it to the reference rather than to the calendar: review when the working gold sample has been out of the drawer a set number of times, because handling ages a sample faster than time does. A heavily used reference may need replacing long before the twelve months are up.

Frequently Asked Questions

People Also Ask